The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection via the extract command.
Ingestion points: The extract subcommand in all CLI scripts (scripts/anysearch_cli.py, scripts/anysearch_cli.js, scripts/anysearch_cli.sh, scripts/anysearch_cli.ps1) fetches and returns external web content to the agent.
Boundary markers: The SKILL.md file includes a specific warning for the agent to treat page content as untrusted data and not to follow embedded instructions.
Capability inventory: The skill is capable of network operations (API search and extraction) and file writing (updating .env and runtime.conf).
Sanitization: Content is returned to the agent as raw Markdown without specific filtering or sanitization of potential injection strings.
[COMMAND_EXECUTION]: The scripts/test_cli.py script uses subprocess.run to execute the skill's CLI tools for contract testing, which involves spawning subprocesses to verify script behavior.
[DATA_EXFILTRATION]: By design, the skill transmits search queries, URLs, and API keys to the vendor's API at api.anysearch.com during normal operation.

anysearch